The Development of Models for Prediction the Behavior of Stocks, Currencies and Commodities Using Machine Learning Techniques

This paper explores the advantages and disadvantages of machine learning methods for prediction the behavior of financial markets. Methods for clustering, classification and regression at the present stage of development in computer technologies has led to a sharp increase in attention to this topic in various fields of human knowledge. The role of financial markets in the development of the world's GDP has increased significantly. People have begun to use of machine learning more actively and wider for building effective investment strategies. The goal of the final qualifying work is to study the application of machine learning methods for forecasting financial markets, as well as the development of models applicable in practical mechanisms of investment portfolio management and exchange trade. To achieve the goal of the work, the following tasks were set and solved: 1) Identify the main components of the concept of machine learning; 2) To categorize the methods of machine learning in relation to financial markets; 3) Select the most effective methods and models of machine learning for solving the problems of price forecasting of financial instruments, optimal construction of the investment portfolio; 4) Carry out a simulation of the dynamics of foreign exchange instruments on the Moscow stock exchange and foreign exchange and over-the-counter platforms; 5) Carry out practical modeling of the investment portfolio in the stock market using machine learning methods; Models for managing investment portfolios were developed using machine learning techniques. In this paper the methods of regression, clustering and classification will be examined the first. Further, this paper presents key aspects of machine learning applications for texts. It includes machine-readable news and press releases issuers, the central banks and infrastructure organizations.